Fig. 4: Utilizing differential processing method to process and learn visual information.
From: Memristor-based adaptive neuromorphic perception in unstructured environments
a The diagram of visual information differential processing. It is important to note that the hardware components, indicated by symbols such as â‘ , â‘¡, etc., correspond to the stages of this processing pipeline. b The experimental design. c The corresponding real-world scenes of the example point visual information processing. d The visual information processing within the yellow box in Fig. 4c, including the change in light intensity (from 0 to 2.56), the differentiation voltage, and the memristor state changes (1 represents high resistance state, 0 represents low resistance state). e Sudden movement of pedestrian at night. f Sudden movement of pedestrian during the day. g Sudden movement of pedestrian hiding behind a car, with background vehicles moving slowly. h The amplitude spectrum comparison between the compressed image and the differentiated image at the last moment in Fig. 4e. i The amplitude spectrum comparison between the compressed image and the differentiated image at the last moment in Fig. 4f. j The amplitude spectrum comparison between the compressed image and the differentiated image at the last moment in Fig. 4g.
